Comment
Author: Admin | 2025-04-28
PHANTOMFor starters, PHANTOM assumes a malicious mining coalition is not the majority, i.e. an attacker has mining power. This puts the asymptotic maximum security threshold of the system at 1/2, similar to that of GHOST and SPECTRE. The PHANTOM protocol differs from SPECTRE in that it enforces a strict ordering over blocks and thus transactions in the system. PHANTOM is usable for smart contract systems in this light but only at the cost of forcing a proper ordering over blocks. The tradeoff can be quantified in the time that it takes nodes in the PHANTOM protocol to reach consensus, which although scalable is not as fast as a protocol like SPECTRE which can run without these guarantees.MiningMining is similar to that in SPECTRE, we quote our description with small changes from the previous article:The mining protocol utilized by PHANTOM follows a similar Proof of Work system as Bitcoin, where the computational puzzles are finding hashes under a target difficulty. At each step in an individual node’s mining process, it examines its view of the blockDAG network and does the following:Finds the set of all blocks with 0 in-degree, denoted B.Computes hashes until it finds a hash h.Creates a block b with hash h, includes B in the header (directed edges to those blocks), and broadcasts b.ConsensusPHANTOM uses purely topological tools for achieving consensus. Differing from the voting scheme used by SPECTRE, the PHANTOM protocol actually determines a “correct blockchain” within the blockDAG as it builds the set of valid blocks in aggregate. Finding this chain happens recursively and forces the total ordering on the transactions included within such a chain. PHANTOM adds these blocks using a greedy approximation algorithm to an optimization problem that will be defined below. For some intuition into this process, the picture below will be useful.The main task at the heart of finding the best, honest blocks begins with finding the maximum k-cluster subDAG indicated by the picture above. The formal problem is defined below. The task of picking the best parameter k also presents an interesting problem, since it involves various tradeoffs given the actual network delay is unknown. For starters, the parameter is closely tied to the expected propagation delay of the entire network. Since we operate under the partial synchronous model, this delay is bounded but not explicitly known.Let’s parse apart this problem statement. Recall from our previous article the definitions of past(x)
Add Comment